Approximations for explanations of inconsistency in partially known multi-context systems

Abstract

Multi-context systems are a formalism to interlink decentralized and heterogeneous knowledge based systems (contexts), which interact via possibly nonmonotonic bridge rules. Inconsistency is a major problem, as it renders such systems useless.In applications involving confidentiality or trust, it is likely that complete knowledge about all system parts is unavailable. To address inconsistency in such scenarios, we extend existing notions for characterizing inconsistency in multi-context systems: we propose a representation of partial knowledge, and introduce a formalismfor approximating reasons of inconsistency.We also discuss query selection strategies for improving approximations in situations where a limited number of queries can be posed to a partially known context. © 2011 Springer-Verlag Berlin Heidelberg.